Digital Recorder Sampling Frequency

[EvanP]

I'm learning to use my new Sony M-10 recorder. It can record in a variety of sampling frequencies and bits/bit rates, and I'm looking for guidance on the optimum setting. So far I've recorded in the default LPCM 44.1kHz/16bit, which seems to have great quality. Michi and others seem to have a preference for the higher bit rate MP3 formats (>44.1kHz/128 kbps), which are definitely attractive because of their smaller file sizes than the LPCM.

Has anyone compared the higher LPCM sampling rates (48 kHz or 96 kHz)? Is anything above Redbook (CD quality) a waste using the built-in mics for field recording?

My application is recording drumming/dance classes for my personal use.

I apologize if this has been covered elsewhere, but a search didn't turn up answers to my questions.

Thanks,
Evan

[michi]

Hi Evan,

the sampling rate determines the frequency range. There something in information theory known as the "Nyquist Limit". It states that the highest frequency that can be captured by taking samples at n Hz is n/2 Hz.

So, 44.1kHz sampling rate means that the highest frequency that can be encoded is 22.05kHz, and 96kHz means that the highest frequency that can be encoded is 48kHz.

On the face of it, the higher sampling rate seems pointless, given that the human ear (and only a very young one, at that) can only hear frequencies up to 20kHz. The reason for sampling at higher rates is to make anti-aliasing easier. Basically, to prevent audio artifacts, after analog-to-digital conversion, the analog signal must be fed through a low-pass filter that removes the sampling noise above 20kHz. With a 44.1kHz sampling rate, you have to use quite a steep (1-octave) filter to do this. The problem is that such a steep filter itself introduces artifacts into the audible parts of the signal. By using a higher sampling rate, the sampling noise is further away from the 20kHz audio limit, so it can be removed with a flatter filter that introduces fewer artifacts.

Another advantage of the higher sampling rate is that it increases effective resolution and reduces the noise floor of the final audio signal a little.

Will any of this make a difference in your situation? Almost certainly not

Basically, for mere mortals with a stereo system for mere mortals, a 320kbit/sec MP3 recording is indistinguishable from a CD. You need to be young, have trained hearing, and have a stereo system that costs in the multiple tens of thousands of dollars to be able to detect the difference between a 44.1kHz/16bit recording and a 96kHz/24bit recording. If you are 30 years or older, chances are you could not tell any difference even with the world's best playback equipment. For field recordings, where you cannot control acoustic conditions and you usually have lots of ambient noise, there is even less point in these higher-quality settings.

Personally, I record at 320kbit/sec, which gives me outstanding sound quality while keeping the file size reasonable. In turn, this means that I run out of memory on the recorder much later, that the files transfer to my computer more quickly, and that the size of my backups is smaller. And I can fit far more of the recordings on my iPhone, where storage space is at a premium.

Cheers,

Michi.

[EvanP]

Michi,
Thanks! I'm aware of the Nyquist rate, but in my experience there is a difference between high resolution audio (analog LPs played on a very good turntable through a very good system, and SACD) and redbook CDs, even though there "shouldn't be". I'm well over 30, and can no longer hear ultrasonic alarm systems in jewelry stores like I used to, but I can still "feel" the difference between LPs and CDs and MP3s.

The only reason to record above MP3 quality for my application is to pick up a little more swing or groove feeling--the stuff that puts a smile on your face. I've recorded two classes so far, the first with 44.1/16 MP3, the second with 44.1/16 PCM. The second class gets me swaying more than the first. Of course, it could totally be in my head.

Do you have experience with how 320kbs compares to some of the PCMs? Do you see any advantage in recording with PCM rather than MP3?

Thanks again,
Evan

[michi]

The only reason to record above MP3 quality for my application is to pick up a little more swing or groove feeling--the stuff that puts a smile on your face. I've recorded two classes so far, the first with 44.1/16 MP3, the second with 44.1/16 PCM. The second class gets me swaying more than the first. Of course, it could totally be in my head.

It could To be sure, you'd have to do double-blind tests. Without that, there is a real danger of perception bias falsifying the results.

Do you have experience with how 320kbs compares to some of the PCMs? Do you see any advantage in recording with PCM rather than MP3?

A few years ago, I did a lot of listening with very high quality (Stax) headphones hooked up to a very high-end stereo and compared a bunch of MP3 rips with their CD originals. For a small number of passages, I believe I could pick up very minor differences. Not that one would sound worse or better than the other; just that there were some differences in sound character. Very minor, almost impossible to describe, and maybe artifacts of my mind rather than real sound differences.

I also found a bunch of recordings that specifically were created to illustrate the artifacts that can be introduced by MP3 recordings, such as pre-echo and post-echo. Those artifacts were very noticeable at lower bit rates but, for 320kbit MP3s, were all but non-existent (to my ears).

I still believe that distinguishing a 320kbit MP3 recording from a CD is impossible for 99.something% of the population, especially when the recording is heard in isolation. In a side-by-side blind comparison, still fewer than 1% of the population would hear any difference, and far fewer would be able to identify which recording is the CD original and which one is the MP3.

Whether it's worth it for you record in CD quality or MP3 quality is really your personal choice. For me, the CD quality isn't worth the cost in memory and inconvenience.

Also keep in mind how you use these recordings. Do you listen to them on a high-quality stereo with full concentration as you might with a great recording of a symphony? Or do you listen to them as a memory aid, to refer back to a workshop you attended and check whether you have all the parts of a rhythm correct? If the latter, there is certainly no point in recording at CD quality.

Cheers,

Michi.

[EvanP]

Michi,
Thanks again. I just got back from an awesome workshop with Manimou Camara. I recorded it at 320 kps MP3 per your suggestion. It sounds fine through PC speakers, but I've just been editing and haven't plugged in my headphones and played along yet. Regardless you're absolutely right--I'm not doing critical listening, but using it as a learning aid. I'm amazed how much more fun even doing warm ups are when playing along with a recording as opposed to a metronome or nothing at all.

I listen to recordings through my Etymotics ear buds, which are pretty revealing, and tend to make you tap your feet to even non-musical CDs.

I am intrigued, however, about how the SACD format with its 1-bit 2.4 MHz samplings sounds so great and conveys so much more emotion than CD-quality recordings. I'm sure I'll play around with different settings over time.

Evan

[Carl]

Geeking out here...

One reason you might consider a higher sample rate is for editing.

I am in the middle of getting my minidisk recordings into my computer (using the D50) this includes normalizing just about everything. With a higher sampling rate (double that of the minidisk recording) I get a slightly cleaner signal, especially during the talk before or after each rhythm. This is where the strongest normalization would be (the strongest amplification).

The theory works like this: having a doubled sampling rate minimizes the artifacting of re-sampling (sampling a sample instead of sampling an analog signal) Less noise from the re-sampling means less noise to be amplified when normalising.

With the new recorder there is this nice feature which records 2 levels at once (with a 20db difference in the recording) and the unit decides which one to keep based on a simple clipping algorithm.

I consider the talking on my recordings as more important than the drumming, mostly because a medium to bad recording of drums is still quite useful, where-as a bad recording of speech is quite frustrating!

C

[michi]

I am intrigued, however, about how the SACD format with its 1-bit 2.4 MHz samplings sounds so great and conveys so much more emotion than CD-quality recordings.

I listened to SACD only once, maybe five or six years ago. On that occasion, I couldn't hear any significant improvement. (This was with an OK stereo system; not bad, but not outstanding either.) But, with only about 6000 titles in existence, SACD is still very much a still-born child.

There is also some evidence to suggest that the extra resolution doesn't really improve things. Here is a passage from the Wikipedia article on SACD:

In the audiophile community, the sound from the SACD format is thought to be significantly better compared to older format Red Book CD recordings. However, In September 2007, the Audio Engineering Society published the results of a year-long trial in which a range of subjects including professional recording engineers were asked to discern the difference between SACD and compact disc audio (44.1 kHz/16 bit) under double blind test conditions. Out of 554 trials, there were 276 correct answers, a 49.8% success rate corresponding almost exactly to the 50% that would have been expected by chance guessing alone.

I'm sure that this isn't the final or only word on the matter but, as the authors of the study say, it does shift the burden of proof to the people who claim SACD is superior.

Cheers,

Michi.

[EvanP]

Michi,
All I know is that on my system (and others I've listened to), SACDs evoke more emotion than CDs, as do LPs on a good turntable. I can't discern differences in frequency between the formats (actually, some LPs are decidedly less "Hi Fi" than either format), but the emotion somehow comes through. Maybe it's psychological suggestion, maybe it's pixie dust, maybe it's snake oil, but it works for me. I know sampling theory, used to religiously read Stereo Review (which basically claimed if you can't measure it, it doesn't exist), and yet my emotions tell me there is an appreciable difference.

Now if I could just find some djembe LPs in good condition!

Until then, I'll keep practicing and recording and practicing...

Besides, live music always sounds best!

Cheers,
Evan

[michi]

Michi,
All I know is that on my system (and others I've listened to), SACDs evoke more emotion than CDs, as do LPs on a good turntable.

I have a number of albums on both CD and LP, and some of those sound better to me on LP too. Remastering can have a lot to with this: the CD version is rarely made from the same master as the LP, so it really is a different recording. As I said, I can't comment on SACD because I don't have experience with it.

Now if I could just find some djembe LPs in good condition!

I don't like your chances

Besides, live music always sounds best!

Sure does! As far as sampling rate is concerned, I would use whichever sounds best to you. After all, that's the point of the whole exercise

Cheers,

Michi.